The final issue of the Journal of Neuroscience for 2008 brings two papers that put a new twist on current thinking about amyloid-β (Aβ). The first, from the lab of Ottavio Arancio at Columbia University in New York, presents evidence that very low concentrations of Aβ, in the range that might be released during normal neuronal activity, promote the synaptic changes associated with learning and improve memory in mice. That is precisely the opposite of what happens with high concentrations of Aβ, which block synaptic strengthening and memory function (see ARF related news story and ARF news story). The results suggest a physiological role for released Aβ in synaptic plasticity and memory, and raise questions about the implications of reducing Aβ levels as a therapeutic approach to Alzheimer disease.

The second report adds to a growing body of work on the Mint/X11 family of adaptor proteins and their role in the processing of amyloid precursor protein to release Aβ. The proteins are clearly involved, but different labs have come up with conflicting results about their exact action. In the latest study, Thomas Sudhof, formerly of the University of Texas Southwestern in Dallas and now at Stanford, reports that knocking out the Mint proteins in transgenic AD mice delays the age-dependent accumulation of Aβ peptides and plaques, apparently by decreasing APP processing by β-secretase. The work stands beside previous studies that show opposite effects, where Mint knockout mice have increased Aβ production, and that Mint proteins regulate γ-secretase. The Mint proteins most likely regulate Aβ production by affecting APP trafficking, but clearly much remains to be learned about the interaction.

Aβ in Tiny Doses
Aβ, produced in high amounts in AD, forms soluble oligomers that at nanomolar concentrations impair synaptic long-term potentiation and memory in mice. Aβ is present in normal brain at much lower levels, and Arancio and colleagues looked at the effects of nearer physiological levels on activity-dependent synaptic plasticity and memory in mice. Lead author Daniela Puzzo, who is also at the University of Catania in Italy, first looked at the effects of Aβ on hippocampal slices during induction of LTP. As previously reported, Aβ at 200 nM elicited impairment of LTP. Dialing back the concentration of Aβ revealed a reversal of that action such that at 200 pM Aβ, there was a significant enhancement in LTP. The effect was specific for activity-dependent potentiation and Aβ: 200 pM Aβ did not enhance basal transmission, and a scrambled peptide did not enhance LTP.

To look at the effects of Aβ in vivo, the investigators injected pM concentrations of Aβ bilaterally into the hippocampus of mice, then tested the animals 20 minutes later for spatial learning in the Morris water maze or for contextual association in a fear-conditioning test. In both cases, the mice performed better after Aβ injection.

How does Aβ enhance LTP? Treating hippocampal slices with 200 pM Aβ did not alter postsynaptic AMPA or NMDA glutamate receptor currents, which can be responsible for LTP. Neither did Aβ appear to affect spontaneous release of neurotransmitters. However, Aβ did enhance stimulated neurotransmitter release during prolonged activation (tetanic potentiation). As with other synaptic effects of Aβ, the potentiation depended on α7 nicotinic receptor activation. It was abolished by selective α7 receptor blocker, and was not seen in hippocampal slices from α7 knockout mice. Moreover, the α7 knockout mice showed no enhancement of memory after hippocampal injection of 200 pM Aβ42.

“These findings strongly support a model for Aβ effects in which low concentrations play a novel positive, modulatory role on neurotransmission and memory, whereas high concentrations play the well-known detrimental effect, culminating in dementia,” the authors conclude. Aβ levels have been reported to be regulated by synaptic activity, suggesting a model where the peptide, released during high-frequency stimulation via vesicle exocytosis, then acts through α7 receptors to modulate glutamate neurotransmitter release.

The results suggest that Aβ can positively affect LTP, both in vitro and in vivo, but the active form of Aβ—soluble oligomeric or other—and its concentration at synapses remains to be clarified. Estimates place in vivo concentrations of Aβ in normal brain in the pM range, and the investigators measured similar levels in the mice by ELISA. If Aβ in low concentrations is a pro-memory molecule, this could have ramifications for efforts to treat AD by lowering amyloid load. “Our work does not challenge the amyloid hypothesis,” the authors write. “However, our lack of understanding of the physiologic role of Aβ may present important issues when designing effective and safe approaches to AD therapy.”

Mint/X11 Mystery
Sudhof and colleagues neatly summarize an extensive literature behind Mint/X11 proteins and APP processing in their paper, using a supplementary table to outline 29 previous studies on the subject. Most of the studies, using Mint overexpressing cells, report that the protein suppresses Aβ production. In agreement with the idea that Mint proteins put a brake on Aβ, knocking out Mint1 or Mint2 in mice has been reported to increase amyloidogenic processing of APP (Sano et al., 2006; Saito et al., 2008).

The current study turns that data on its head. Coauthors Angela Ho and Xinran Liu crossed Mint knockout mice into two different AD transgenic mice lines bearing mutated human APP and presenilin genes. In this model, deletion of any of three Mint isoforms delayed early plaque accumulation in hippocampus and cortex, but the magnitude of the effect was isoform-dependent. Mint2 deletion had the greatest impact, and Mint3 the least. A similar pattern was seen for cortical plaques, and for Aβ40 and 42 peptide levels.

The authors attribute the differential effects of the isoforms to their expression pattern in the brain. Staining of brain sections revealed that Mint2 and APP share a similar location in the soma of excitatory pyramidal neurons in the CA3 region of hippocampus. Mint1, on the other hand, did not colocalize with APP. Mint3 is barely expressed in the brain, they write, citing previous work.

To look at the effect of Mint on APP processing, the researchers made neurons lacking all three Mint isoforms. Mice with floxed Mint alleles were crossed with AD mice, and neurons isolated from the resulting animals were infected with a Cre recombinase-expressing virus. These cells made less Aβ40 and 42 compared to Mint expressing cells. The effect on APP processing was to decrease the accumulation of the C-terminal fragments produced by β-secretase and α-secretase. The accumulation of CTFs was only seen in the presence of a γ-secretase inhibitor which prevented their further processing. When the inhibitor was washed out, cleavage of the CTFs proceeded normally, suggesting that deletion of the Mint proteins do not affect γ-secretase activity. This is in contrast to previous results from multiple labs using overexpression, and one report of RNAi knockdown (Xie et al., 2005) that suggested Mint proteins regulate APP processing by γ-secretase.

What do the Mint proteins do? They associate with APP via its intracellular domain, and may play a role in APP trafficking. The knockout mouse phenotype is consistent with an effect on APP trafficking, the authors conclude, because changes in where APP sits in the cell and for how long could influence processing by β-secretase, which occurs intracellularly in endosomes. It will be interesting to see how the details of Mint/APP interactions unfold going into 2009 and beyond.—Pat McCaffrey


  1. This is an important finding in keeping with our studies presented at the neuroscience meeting to demonstrate that the physiological role of amyloid-β protein is to enhance memory. This explains the failure of drugs that totally reduce amyloid-β to enhance memory. Modulators will be necessary to treat Alzheimer disease.

    View all comments by John Morely
  2. Evidence supporting a role for the Mint/X11 proteins as regulators of APP metabolism and Aβ production has been accumulating for 10 years now. Unfortunately, much of the data yield opposing models; Mints appear capable of imparting pro- or anti-amyloidogenic effects. Typically, when this stark a disagreement occurs, the cause is either that we are missing an important part of the puzzle or the process is much more complicated than we envision.

    This latest paper from the Sudhof lab (Ho et al., 2008) sought to resolve the issue by performing an extensive (and impressive) array of assays of APP metabolism in APPswe/PS1dE9 mice deleted for each of the three Mint family members. This is a first-rate paper with very high-quality data that address an important question in the field of Alzheimer’s research. Unfortunately, it does not resolve the differences, but it does provide new data that may help focus the search on the source of the differences and the missing pieces. Two key observations are discussed, and we note one concern.

    They found that deletion of any one of the three Mints (Mint1-3, aka X11; X11L (X11-like); and X11L2 (X11-like2) delayed the progression of plaques in this mouse model and decreased the products of β-site cleavage of APP and the levels of Aβ40 and Aβ42, particularly at earlier times. This interesting observation is surprising for two reasons. First, the three Mints are differentially expressed in the brain, with Mint1 predominantly in interneurons, Mint2 in pyramidal neurons, and Mint3 ubiquitously expressed. Thus, for each of them to have such strong effects on plaque development and APP metabolism requires either that 1) pathologic Aβ generation occurs in multiple cell types, 2) cells expressing submaximal amounts of Mints may be key to amyloidogenesis, or 3) deletion of one Mint can impact the actions of the others. Indeed, some important data are found in Supplementary Table 2, showing that knockouts of Mint1 or Mint2 significantly decrease (by 55 percent and 32 percent, respectively) the expression of the other, while knocking out Mint3 only lowers Mint2 levels (19 percent decrease). It appears that these values represent changes in Mint expression in the whole brain, so these differences would be magnified in the relevant cell types. The other surprising aspect of these data is that they appear to be completely opposite to two studies from the Suzuki laboratory (Sano et al., 2006; Saito et al., 2008), which reported that knockout of Mint1 or Mint2 led to increased β-site cleavage of APP and increased generation of Aβ peptides. The two laboratories also differ in that the Suzuki lab reports no changes in the levels of Mint1 when Mint2 was deleted and vice versa. Ho et al. (Ho et al., 2008) cites strain differences as a possible source of these differences. While the Suzuki lab clearly states C57BL/6 as the predominant background in their studies, it is difficult to determine the strain background used by Ho et al. If such dramatic differences in the effects of Mints on amyloidogenesis are really the result of strain differences, then it may highlight a key finding that could well prove important in humans. Two other issues are viewed as potentially contributing to these apparent differences, though seem unlikely to explain it all. These are the time points studied (Ho et al. look at three time points of six, nine, and 12 months and noted several differences in effects at different times, while the Suzuki lab focused on 12 months) and the fact that the Ho et al. work is all performed in mice engineered to express the amyloidogenic human APPswe and PS1dE9 mutants. For example, the Ho et al. study found that when Mint3 was knocked out the levels of Aβ in both the cortex and the hippocampus were decreased at three months but increased at the 12-month time point (the same trend and time point observed in Saito et al., 2008).

    The concern we note is an issue with both the Sudhof and Suzuki lab publications, and that is the general underappreciation of the potential role of Mint3 in amyloidogenesis. Because Mint1 and Mint2 are expressed in a neuronal-specific fashion, researchers have consistently minimized or ignored the ubiquitously expressed Mint3. However, Mint3 is present in human brain and in an expression pattern reflecting that of Mint2 (Shrivastava-Ranjan et al., 2008) and APP. Ho et al. (Ho et al., 2008) clearly document a strong effect of Mint3 deletion on amyloidogenesis, particularly at the critical early time points, and on Mint2 expression levels. All three Mint proteins share the central and APP-binding PTB domain and C-terminal, dual PDZ domains that also are required for their recruitment to membranes through direct binding to ARF GTPases. They differ in their N-termini, with Mint3 lacking any additional domains, Mint1 and Mint2 having a Munc18 interacting domain (from where the name Munc18-interacting proteins is derived), and only Mint1 having a CASK binding domain, making it the sole member of the Mint family found in the heterotrimeric complex of Mint1/CASK/Velas. The observation in Ho et al. that CASK knockout mice do not share the phenotypes they report for the Mint1 knockout supports the conclusion that Mint1 also has actions independent of that trimeric complex. Similarly, Mint2 is found both at the cell surface and at the Golgi in cultured cells, while Mint3 appears to be predominantly at the Golgi. We have recently shown that Mint2 and Mint3 are present on the post-Golgi vesicular carriers that transport APP from the Golgi toward the cell surface (Shrivastava-Ranjan et al., 2008). Thus, the data are growing that Mint2 and Mint3 are mediators of APP traffic and have all the features of Arf-dependent adaptors that put them squarely in the same sphere as GGAs, putative mediators of SorLa/LR11 and β-secretase traffic.

    If one is looking for a molecular model that may explain the confusing and inconsistent data from studies of Mints in APP metabolism, it is sufficient to invoke the field of membrane traffic, an equally confusing and discrepant field. As basic scientists we view Alzheimer disease as a membrane traffic disorder whose molecular mechanisms will only be understood in the context of the normal regulation of sorting and residency of APP and other cargos (e.g., secretases and interactors like LR11) as they move throughout the endomembrane system. We strongly encourage members of the two labs to exchange their valuable animals and pursue these questions in hopes of either identifying a previously unknown piece of this puzzle or unraveling this increasingly complex but critical mechanism.


    . Deletion of Mint proteins decreases amyloid production in transgenic mouse models of Alzheimer's disease. J Neurosci. 2008 Dec 31;28(53):14392-400. PubMed.

    . X11 proteins regulate the translocation of amyloid beta-protein precursor (APP) into detergent-resistant membrane and suppress the amyloidogenic cleavage of APP by beta-site-cleaving enzyme in brain. J Biol Chem. 2008 Dec 19;283(51):35763-71. PubMed.

    . Enhanced amyloidogenic metabolism of the amyloid beta-protein precursor in the X11L-deficient mouse brain. J Biol Chem. 2006 Dec 8;281(49):37853-60. PubMed.

    . Mint3/X11gamma is an ADP-ribosylation factor-dependent adaptor that regulates the traffic of the Alzheimer's Precursor protein from the trans-Golgi network. Mol Biol Cell. 2008 Jan;19(1):51-64. PubMed.

    View all comments by Amanda Caster
  3. This is a very good study and I agree wholeheartedly with the results. It again shows that amyloid-β (Aβ) peptides do have a normal physiological role in the brain. Work from our group over the past few years has shown that Aβ42, in particular, is present in, and on, neurons in normal brains, and that the binding of exogenous Aβ42 to neuronal surfaces, including synapses, is mediated through the high affinity binding of Aβ42 to the α7 nicotinic acetylcholine receptor.

    This new work by Puzzo et al. again suggests a normal role for Aβ peptides in neurons, perhaps at the level of the synapse, that Aβ peptides are not toxic, and that the “purpose” of Aβ peptides and the enzymes that produce them, α- and γ-secretase, is not just to give us Alzheimer disease. Interestingly, it seems that the only difference between comparable neurons in normal and AD brains is that that latter seem to accumulate large quantities of Aβ sequestered in their lysosomal compartments, and many of these eventually die and undergo lysis to form a debris cloud referred to as classical, dense-core amyloid plaques. This study also emphasizes that exogenous Aβ has selective affinity for α7-rich regions of the brain parenchyma, and some of our more recent work has shown that the blood can serve as a chronic source of this Aβ in brain regions experiencing local breakdown of the blood-brain barrier.

    In short, I believe that these investigators are on the right track. Now the trick will be to get the rest of the field to take notice, stop “gnawing at the ends of old plots,” and finally allow fresh ideas to infiltrate the field that will lead to resolving this perplexing and devastating disease. Thank you for this refreshing electrophysiological study.

    View all comments by Michael R D'Andrea

Make a Comment

To make a comment you must login or register.


News Citations

  1. Earliest Amyloid Aggregates Fingered As Culprits, Disrupt Synapse Function in Rats
  2. “Natural” Aβ Oligomers Cause Transitory Cognitive Disruptions

Paper Citations

  1. . Enhanced amyloidogenic metabolism of the amyloid beta-protein precursor in the X11L-deficient mouse brain. J Biol Chem. 2006 Dec 8;281(49):37853-60. PubMed.
  2. . X11 proteins regulate the translocation of amyloid beta-protein precursor (APP) into detergent-resistant membrane and suppress the amyloidogenic cleavage of APP by beta-site-cleaving enzyme in brain. J Biol Chem. 2008 Dec 19;283(51):35763-71. PubMed.
  3. . RNA interference-mediated silencing of X11alpha and X11beta attenuates amyloid beta-protein levels via differential effects on beta-amyloid precursor protein processing. J Biol Chem. 2005 Apr 15;280(15):15413-21. PubMed.

Further Reading

Primary Papers

  1. . Picomolar amyloid-beta positively modulates synaptic plasticity and memory in hippocampus. J Neurosci. 2008 Dec 31;28(53):14537-45. PubMed.
  2. . Deletion of Mint proteins decreases amyloid production in transgenic mouse models of Alzheimer's disease. J Neurosci. 2008 Dec 31;28(53):14392-400. PubMed.